1. Introduction

High energy, ultra-short laser pulses are a very powerful tool in the fields of material science, biophysics, and molecular science. In addition to pulse energy and duration, repetition rate is a critically important parameter for lasers used in these fields as it determines the rate of data acquisition in experiments and, consequently, the volume of data acquired. As a means of producing high energy, high repetition rate, ultra-short pulses, OPCPA (optical parametric chirped pulse amplification) systems are being developed worldwide. In comparison to standard Ti:sapphire laser systems, OPCPA offers larger versatility in achievable pulse duration and central wavelength, due to the exceptionally large phase matching bandwidth in non-collinear configurations [1], and high repetition rates due to the negligible heat load in the crystal [2, 3]. OPCPA with picosecond pulses offers additional advantages, such as higher contrast, broad spectral bandwidth, and the ability to use thin nonlinear crystals [4]. OPCPA using pulses on the scale of 1 ps puts strict requirements not only on precise synchronization between signal and pump pulses, but also on the parameters of the pump lasers themselves.

The key characteristics of a suitable pump laser for picosecond OPCPA are high energy at repetition rates of one kHz or higher, good beam quality, good long and short term energy stability, and picosecond pulse length. There are multiple approaches to produce such laser pulse trains; arguably the most common technologies are fiber amplification [5,6] and coherent combination of fiber amplifiers [7], Innoslab amplification [8], and thin disk amplification; however, other technologies are also being developed [9]. Currently, thin disk based amplifiers are able to produce picosecond pulses with the highest energies and highest beam quality at ≥1 kHz repetition rates.

Thin disk lasers were introduced in 1994 by the group of Adolf Giesen at the University of Stuttgart [10]. As the technology matured, thin disk laser heads started to be used not only in CW lasers, but also in regenerative amplifiers [11] and mode-locked oscillators [12, 13]. The factors that limit the output pulse energy in practice are mainly the damage threshold of optical components and nonlinear effects such as self-phase modulation or self-focusing. When high energy goes together with high repetition rate, the amplifiers must be also able to deal with high average power and heat dissipation. Significant steps toward high picosecond pulse energies (>10 mJ) and repetition rates (≥1 kHz) have been made with thin disk regenerative amplifiers by a number of groups in recent years [14–16 ]. Additionally, synchronization of thin disk regenerative amplifiers to OPCPA seed pulses on the femtosecond scale has been experimentally demonstrated [17,18].

In this paper we report on recent results of a high energy frequency doubled thin disk regenerative amplifier. Our system generates 42 mJ picosecond-level pulses at 515 nm with a repetition rate of 1 kHz. The thin disk regenerative amplifier at the heart of the laser system is able to generate 1.2 ps, 90 mJ pulses (104 mJ uncompressed) at 1030 nm, which is one of the highest reported pulse energies for a picosecond-level kHz laser. The laser has been constructed to serve as one of the pump lasers of an OPCPA chain under construction at the ELI-Beamlines facility. As this is intended to be part of a larger beamline for a user facility, the emphasis is placed less on achieving the highest performance parameters possible and more on stability, reliability, and integrability into the larger laser system.

2. Experimental setup

The 1030 nm seed pulses for the regenerative amplifier are derived from the long-wavelength portion of the spectrum of an ultra-broadband modelocked Ti:Sapphire oscillator with an 80 MHz repetition rate (Femtolasers, Rainbow); using a single laser to seed the OPCPA pump lasers and the OPCPA itself assures initial synchronization between the broadband signal pulses and pump pulses [14,17]. Amplification of the 1030 nm pulses in the regenerative amplifier significantly reduces the bandwidth of the laser pulses which has two effects: it broadens the final compressed pulse duration and shortens the duration of the chirped pulse in the amplifier cavity, leading to higher B-integral. To reduce these negative effects we try to maintain the broadest possible spectral bandwidth of the amplified pump pulses via pre-amplification in fiber, which has a comparatively large amplification bandwidth.

The 1030 nm seed pulses are coupled into a polarization maintaining fiber and stretched in a chirped fiber Bragg grating (CFBG) stretcher (Teraxion). The stretching factor of the CFBG (−478 ps/nm), as well as higher orders of dispersion, are matched to the grating compressor following the regenerative amplifier. The CFBG has a reflective bandwidth of roughly 4 nm giving an input pulse duration of ~2 ns. Subsequently, pulses are amplified to 14 µJ by a chain of Yb-doped fiber amplifiers, the repetition rate is reduced to 1 kHz with a KD*P Pockels cell and the pulses are seeded into the thin disk regenerative amplifier.

The regenerative amplifier contains two laser heads (TRUMPF TruMicro series 5000), each with an Yb:YAG thin disk glued to a water-cooled diamond substrate, which assures efficient heat removal and minimal thermal lensing. The disks have a thickness of roughly 1/10 mm, diameter of 9 mm, and a doping level of ~10 at. %. Each disk is pumped by 500W at 969 nm from a wavelength stabilized laser diode module. The pump light is delivered to the head via fiber. The laser head, consisting of a collimator, parabolic mirror and prisms, images the pump light 18-times on the disk, resulting in nearly complete absorption of the pump light. The pump spot on the disk is slightly blurred to reduce the thermal gradient on the edges and also to support a larger cavity mode size. The pump radiation at 969 nm matches the 1 nm-wide zero phonon absorption line, and hence the thermal load caused by the quantum defect is reduced by 34% in comparison to the conventional pump wavelength of 940 nm. The lower temperature helps to reduce stress in disks and also to increase the gain, as the quasi-three-level nature of the Yb:YAG crystal means lower reabsorption losses of 1030 nm photons at lower temperatures. In order to further decrease the thermal load on disks and also increase the efficiency of the regenerative amplifier, power supplies of the diode modules are operated in a pulsed regime with 50% duty cycle at 1 kHz. A simplified layout of the regenerative amplifier is shown in Fig. 1.

 

Fig. 1 Layout of thin disk laser system. YDFA: ytterbium doped fiber amplifier, PP: KD*P pulse picker, PC: BBO Pockels cell, ISO: optical isolator, ROT: Faraday rotator.

Download Full Size | PDF

The Pockels cell in the laser cavity consists of a 12×12×20 mm³ BBO crystal held in a water-cooled holder with copper electrodes. The cavity of the regenerative amplifier is about 5 m long and is carefully designed to have the mode as large as possible at the Pockels cell crystal (4.5 mm, 1/e ² diameter) while also matched optimally to the size of the pump spots on the disks (4 – 4.2 mm, 1/e ² diameter). The design also avoids any foci in the cavity to keep the B-integral below π. A more detailed description of cavity design considerations for a similar, smaller system can be found in [19].

After amplification, the pulses are coupled out of the cavity by the Pockels cell, and resized with a telescope to match the target fluence desired on the SHG crystal. The pulses are then compressed in a Treacy compressor consisting of a multi-layer dielectric (MLD) grating pair (Plymouth Grating Laboratory). The gratings have a groove density of 1740 lines/mm, dimensions of 84 mm × 96 mm and 70 mm × 150 mm, and are separated by a distance of 4 m with an incidence angle of 62°. The compressor is followed by a 1.47 mm thick LBO crystal, where the second harmonic at 515 nm is generated. This green light is later used as a pump for a single OPCPA stage.

3. Performance of the laser system

In the above described regenerative amplifier, pulses have been amplified up to 104 mJ at 1 kHz, where the optical-to-optical efficiency reaches 31% before compression. Up to that level, the amplification is linear with respect to pump energy and the only limitation in energy is given by the damage threshold of optical components (see Fig. 2(a)). The number of round trips in the cavity is chosen so that an amplified pulse at the given pump energy just reaches saturation. Therefore the amplifier is not sensitive to seed energy fluctuations and its energy stability measured during 1 hour has a standard deviation of better than 0.8% (measured after compression), as shown in Fig. 2(b).

 

Fig. 2 a) Output energy and optical-to-optical efficiency of the regenerative amplifier b) Energy stability measurement of the compressed beam (90 mJ after losses in the compressor) during 60 minutes of operation, standard deviation σ < 0.8%.

Download Full Size | PDF

The spatial quality of the output beam is high with M ² of 1.04 and 1.2 in the x and y axes, respectively. During a ramp-up of energy we observe a small increase of M ² in the y axis while in the x axis it remains practically unchanged. We suspect that this is due to the nonzero incidence angle on the disk in the horizontal plane, which results in a longer path length through the crystal in that plane. After compression, the M ² remains the same and no noticeable degradation of beam quality is observed. The M ² measurement is shown in Fig. 3 as well as the nearly diffraction limited beam profile of the compressed beam.

 

Fig. 3 M ² measurement of amplified beam at 104 mJ (left); beam profile of a compressed beam at maximum energy (right).

Download Full Size | PDF

The pulses are compressed to 1.2 ps in the MLD compressor, which is 1.2 times the transform limit given the pulse bandwidth of 1.6 nm. The average diffraction efficiency of each grating is about 96.5%, giving an overall compressor efficiency of nearly 87%. An auto-correlation trace of output pulses is shown in Fig. 4. Visible wings on the sides of the trace originate from uncompensated dispersion of the CFBG stretcher. Since the stretcher is sensitive to temperature, dispersion tuning by local temperature change will be implemented in the future.

 

Fig. 4 Auto-correlation trace of compressed pulses with energy of 90 mJ.

Download Full Size | PDF

While the pulse duration, long term stability, and spatial beam quality described above are all measured while the regenerative amplifier operates at 104 mJ, the amplifier output is limited to a very conservative 80 mJ in practice. Because the laser will be integrated into a larger beamline intended for users, we limit the output energy to ensure long-term reliability and durability of the laser system and to remain comfortably below the damage threshold of the resonator optics. After limiting the output energy of the regenerative amplifier to 80 mJ (69 mJ after compression) the pulses are frequency doubled in the LBO crystal, resulting in 42 mJ at 515 nm (61% conversion efficiency).

4. Conclusion

We have reported on the generation of 90 mJ, 1.2 ps pulses with a repetition rate of 1 kHz delivered in a near-fundamental spatial mode at a wavelength of 1030 nm with an energy fluctuation of less than 0.8% from a Yb:YAG thin disk chirped pulse regenerative amplifier. Second harmonic of 69 mJ at 1030 nm yields 42 mJ in green which is, to our knowledge, the highest energy at or near this wavelength at a kilohertz repetition rate and picosecond-level pulse duration. The system presented here is a reliable pump laser for a picosecond OPCPA thanks to its high stability and excellent beam quality.